As recently described by Geurts (2023), the study of the notion of ‘common ground’ in pragmatics takes face-to-face conversations as a model for communication. In face-to-face settings, interlocutors typically have (more or less) simultaneous attitudes concerning conversational updates: the moment that the speaker states that p, the addressee will immediately also come to believe/accept that p (or at least that p was stated). Likewise, formal characterizations of common ground on offer, (e.g., in terms of common belief/acceptance/knowledge/commitment, see e.g., Stalnaker (2014); Clark and Marshall (1981); Geurts (2019)) typically assume simultaneous attitudes. At best, these definitions do
not say anything about when interlocutors have the relevant attitudes.

However, the concept of common ground has, without much hesitation, (except by Harris (2020); Semeijn (2024)) been extended to asynchronous non-face-to-face conversations in which speaker and addressee do not have simultaneous attitudes. For instance, several authors in philosophy/semantics of fiction (e.g., Zucchi (2021); Eckardt (2014); Maier and Semeijn (2021)) use the notion of common ground between an author (e.g., Austen) and her audience (e.g., myself) to analyze fictional discourse and/or truth in fiction.

The intuitive justification for this move is simple: Non-face-to-face conversations work in essentially the same way as face-to-face conversations. Consider a simple case of asynchronous communication: A man on his deathbed writes a letter to his daughter that states “The medallion is in Nouvion” (p) and dies. The daughter reads her father’s ‘letter from beyond the grave’ and comes to believe that p. Similar to cases of face-to-face communication, the notion of ‘common ground’ has a dual function (see Geurts (2023)) to play in our explanation of this communicative act: First, in writing his letter, father proposed
to update the common ground between him and his daughter with p. Second, father was able to felicitously phrase his contribution in this way (i.e., using the definite description “the medallion”) because it was already common ground between him and his daughter that there was some specific medallion.

As natural as that may sound, if we assume straightforward time-indexing (i.e., the kind of time-indexing that makes perfect sense for face-to-face settings) for standard ‘mentalistic’ characterizations of common ground (e.g., in terms of common belief), then p is never actually common ground between father and daughter. Below B_x^t_nϕ means x believes that ϕ at tn:

p is common ground at t1 between speaker a and hearer b iff

B_a^t₁p
B_a^t₁B_b^t₁p
B_a^t₁B_b^t₁B_a^t₁p
...

B_b^t1p
B_b^t₁B_a^t₁p
B_b^t₁B_a^t₁B_b^t₁p
...

By the time that the daughter comes to believe that p, the father has already passed away.

Multi-agent system logics may aid philosophers of language here: Loosely following Halpern and Moses’s (1990) concept of ‘eventual common knowledge’, we might propose the following definition of ‘eventual common ground’:

p is common ground between speaker a and hearer b iff

∃tB_a^tp
∃tB_a^t∃t'B_b^t'p
∃tB_a^t∃t'B_b^t'∃t''B_a^t''p
...

∃tB_b^tp
∃tB_b^t∃t'B_a^t'p
∃tB_b^t∃t'B_a^t'∃t''B_b^t''p
...

Now p is common ground between father and daughter: They both believe that p at some point in time, both believe at some point in time that the other believes that p at some (possibly other) point in time, etc.). However, this notion also will not do. On this definition it is always common ground between father and daughter that p. However, we require a notion of common ground that is dynamic: Common ground is something that can grow as communication progresses. Intuitively, p became common ground at some point thanks to father’s letter.

I propose that we require a shift from defining ‘common ground between agents’ at a certain time to defining common ground between ‘agents at a certain time’:

p is common ground between speaker a at t₁ and hearer b at t₂ iff:

B_a^t₁p
B_a^t₁∃t(t≥t₁∧B_b^tp)
B_a^t₁∃t(t≥t₁∧B_b^t∃t'(t'≤t∧B_a^t'p))
...

B_b^t₁p
B_b^t₁∃t(t≥t₁∧B_a^tp)
B_b^t₁∃t(t≥t₁∧B_a^t∃t'(t'≤t∧B_b^t'p))
...

The basic idea is that in case father had the relevant beliefs in his time (e.g., at t₁ father believes that p and that his daughter would come to believe that p, etc.) and the daughter had the relevant beliefs in her time (e.g., at t₂ daughter believes that p and that her father used to believe that p), then it is common ground that p between father in his time and daughter in her time. I explore predictions of and potential issues with this definition (e.g., this definition allows us to specify what is common ground between speaker at t₂ and hearer at some earlier time t₁, a situation arguably only possible in fiction).

References

Clark, H. H. and Marshall, C. R. (1981). Definite reference and mutual knowledge. In Joshi, A. K., Webber, B. L., and Sag, I. A., editors, Elements of Discourse Understanding, pages 10–63. Cambridge University Press, Cambridge.

Eckardt, R. (2014). The Semantics of Free Indirect Discourse. Brill Publishers, Leiden.

Geurts, B. (2019). Communication as commitment sharing: speech acts, implicatures,
common ground. Theoretical Linguistics, 45:1–30.

Geurts, B. (2023). Common ground in pragmatics.

Halpern, J. Y. and Moses, Y. (1990). Knowledge and common knowledge in a distributed environment. Journal of the ACM, 37(3):549–587.

Harris, D. W. (2020). We talk to people, not contexts. Philosophical Studies, 177(9):2713–2733.

Maier, E. and Semeijn, M. (2021). Extracting fictional truth from unreliable sources. In Maier, E. and Stokke, A., editors, The Language of Fiction, pages 186–220. Oxford University Press, Oxford.

Semeijn, M. (2024). Common ground: In sensu composito or in sensu diviso. Journal of Philosophical Logic.

Stalnaker, R. (2014). Context. Oxford University Press, Oxford.

Zucchi, A. (2021). On the generation of content. In Maier, E. and Stokke, A., editors, The Language of Fiction, pages 107–130. Oxford University Press, Oxford.

Sam and Maria are walking in a park when Maria slips on a banana peel and falls. Sam laughs. A silence ensues as Maria gets up. Maria is angry. Sam had no intention of conveying any information to Maria by behaving in this way. Nevertheless, by his laughter and lack of apology, Sam implies certain things. In particular, this behavior, combined with the knowledge that Maria and Sam already had in common, allows for the following interpretation:

(p) Sam lacks respect for Maria.[1]

If we ask Sam: “What did you mean by that laugh? And by that silence?”, he could legitimately answer: “Nothing! I didn't mean anything. I didn't want to communicate anything.” We should think of this example in such a way that this answer is neither false nor misleading.

Nevertheless, even if he did not mean to say anything, by his behavior, Sam somehow implies that (p). In other words, by laughing and failing to apologize, Sam involuntarily sends the message that (p). Maria can legitimately reproach him for this and can, moreover, expect Sam to know that she can reproach him for this even before she tells him so. I will say that, with this behavior, Sam means (p) by unintentional permission or, equivalently, that he unintentionally allows his behavior to mean that (p). Indeed, although it is unintentional, with his behavior Sam allows, permits, or legitimizes certain interpretations by Maria, including (p). On the one hand, he had no intention of conveying this message but, on the other hand, he manifestly could have prevented this interpretation. Because he could, he gives Maria reason to interpret him as conveying certain messages, including (p).

In this article, I present an analysis of this phenomenon of meaning by unintentional permission. I will do so by presenting a model inspired by the work of Grice and his heirs that I call “the extended Gricean model”, abbreviated EGM. More specifically, I will defend the following three points:

(1) Theories of implicatures developed by Grice (1989) and his heirs (for a review, see Davis 2024) fall short of being able to analyze meaning by unintentional permission insofar as these theories focus on intentional meaning (or ostensive-inferential communication).

(2) Some adaptations of the Gricean framework allow us to account for these cases. This leads me to present the EGM and define its key notions, which are adapted from Gricean notions: meaning by unintentional permission (which replaces “utterer’s meaning” in the EGM), to allow x to F (which replaces “to intend x to F” in the EGM), and the principle of rationalization of permission (which replaces the Gricean Cooperative Principle and, more generally, the principle of rationalization of Kasher (1982)).

(3) Finally, I argue that the EGM can explain many interpretations that we make on a daily basis that escape the existing (post-)Gricean theories as well as the other communication theories that I know of. We will see in particular how this model explains why, for example, Sam is responsible for what he implies by his behavior, even if he does not mean anything by his behavior. This is because, in a nutshell, he has some control over what he allows his behavior to mean.

To give a foretaste of the EGM, here are the definitions of the key notions mentioned above:

Meaning by unintentional permission

By the production of the stimuli x, the sender S means that p by unintentional permission to the receivers R if, and only if,

S unintentionally allows x

(i) to generate the mental effects e in R – typically the relevant effect is to make p manifest to R – and

(ii) to make (i) mutually manifest to S and R.

Note: Specifying e is tantamount to specifying p, i.e. the content of meaning by involuntary permission.

To allow x to F

The sender S allows the stimuli x produced by S between the instants t0 and t1 to F if and only if

- S has control over whether S Fs between t0 and t1 and

- it is manifest to S between t0 and t1 that S may F with x.

Note: “S has control over whether S Fs” can be analyzed in at least two ways: Either we say that it is within S's power not to F (Chisholm 1967) or F-ing is produced by a mental mechanism of S that is sensitive to reasons (cf. the notion of ‘moderate guidance-control’ (Fischer and Ravizza, 1998, ch. 3)). In any case, S can be taken to be responsible for F-ing.

Principle of rationalization of permission

If S allows in a mutually manifest way x to generate e in R, then one may assume that these effects are evaluated by S as being conducive to S’s goals, ceteris paribus. In particular, one may assume that S would not have allowed x to generate e in R if this had been evaluated as less conducive to S's goals than not allowing it, ceteris paribus.

Note: “to evaluate” is here to be understood as a basic, rapid, non-demanding cognitive process (see Moors, 2022, ch. 6-7).

References

Chisholm, R. M. (1967). He could have done otherwise. The Journal of Philosophy, 64(13), 409–417.

Davis, W. (2024). “Implicature”, The Stanford Encyclopedia of Philosophy, Zalta, E. & Nodelman, U. (eds.), <https://plato.stanford.edu/archives/spr2024/entries/implicature/>.

Fischer, J. M., & Ravizza, M. (1998). Responsibility and Control: A Theory of Moral Responsibility. Cambridge University Press.

Grice, H. P. (1989). Studies in the Way of Words. Harvard University Press.

Kasher, A. (1982). Gricean Inference Revisited. Philosophica, 29(1), 25–44.

Moors, A. (2022). Demystifying emotions. Cambridge University Press.

[1] Or something like that. The precise content of (p) is not relevant to my argument. It should be understood in a flexible enough way to be replaced indifferently by: Sam does not respect Maria at her true value; Sam does not treat her in a sufficiently respectful way; Sam respects her less than she would like; etc.